While I agree with much of JSE's post (recently most of my projects are joint, and I am enjoying them much more than my solo work), I have some reservations about his advice. I think *at the beginning* of your career, you should make sure that a significant percentage of your best papers are solo. Here are some reasons why:

a) If most of your papers are collaborations, the committees  would have to "dig deeper" (to quote Joel David Hamkins). Why make it harder for your supporters to support you? 

b) Once people know that you are productive and creative, collaborations would be easier to come by and even more enjoyable. In fact, you would probably get more credits than you deserve! (while this is a nice problem to have, now you should make sure that the graduate students who write with you get the fair share of credits).   

c) While working solo is harder and more boring, it does train you in different aspects that would be quite useful for your career: for example you have to be more careful with yourself since nobody will be around to point out your mistakes. 

One more thing that is already mentioned by Joel, but I think worth repeating: when you collaborate, make sure you diversify (both in people and topics).